-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsegment.py
224 lines (198 loc) · 8.51 KB
/
segment.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
from typing import List
import cv2
import numpy as np
from c3d_feature import C3D_feature
import torch
import torchvision
import math
def get_dist(x, y, i, j):
return math.sqrt(math.pow(x-i,2)+math.pow(y-j,2))
def get_simi(f1, f2):
return torch.cosine_similarity(torch.from_numpy(f1),torch.from_numpy(f2), dim=0)
class Frame:
def __init__(self, frame:cv2.Mat, idx=-1) -> None:
self.frame = frame
# self.clear_score = self.getClearFeature(frame)
# print(f"clear: {self.clear_score}")
self.symmetry_score = self.getSymmetry(frame, 1)+self.getSymmetry(frame, 0)
# print(f"symmetry:{self.symmetry_score}")
# self.symmetry1_score = self.getSymmetry(frame, 1)
# self.symmetry0_score = self.getSymmetry(frame, 0)
# self.hog = self.getHogFeature(frame)
self.color_score = self.getColor(frame)
# print(f"color: {self.color_score}")
# self.pos = 0 # For aesthetic：最终摘要中的第几帧
# self.belongWhichSegment = 0 # For aesthetic: 属于第几个片段
# self.raw_id = idx # 原始视频中的第几帧，用于实验评分
def getSymmetry(self, img:cv2.Mat, code:int):
img_cal = cv2.resize(img,(270,480))
# 创建对称图
img_flip_src = cv2.flip(img_cal,code)
# 将特征点对称
# 创建sift计算子
feature_point_size = 100
sift = cv2.SIFT_create(feature_point_size)
img_gray = cv2.cvtColor(img_cal, cv2.COLOR_BGR2GRAY)
img_flip_gray = cv2.cvtColor(img_flip_src, cv2.COLOR_BGR2GRAY)
# 获取该图像的特征点
sift_src=sift.detectAndCompute(img_gray,None)
sift_flip=sift.detectAndCompute(img_flip_gray,None)
avg_sc = 0.0
theta = 50.0
for src_id, src_p in enumerate(sift_src[0]):
x,y= src_p.pt[0], src_p.pt[1]
dist = []
simi = []
for flip_id, flip_p in enumerate(sift_flip[0]):
i,j = flip_p.pt[0], flip_p.pt[1]
dis = get_dist(x,y,i,j)
if dis <= theta:
dist.append(dis)
if len(dist) != 0:
dist = torch.log_softmax(torch.Tensor(dist),dim=0)
sc =0-min(dist)
avg_sc += sc/ float(feature_point_size)
return avg_sc
def getColor(self, img:cv2.Mat):
img_src = cv2.resize(img, (25,50))
img_hsv = cv2.cvtColor(img_src,cv2.COLOR_BGR2HSV)
score = 0.0
cnt =0
for i,row in enumerate(img_hsv):
for j,pixel in enumerate(row):
cnt+=1
sc = (float(pixel[1])*0.5+float(pixel[2]))*0.5/255.0
score+= sc*100.0
return score/(25.0*50.0)
def getClearFeature(self, src):
cal_img = cv2.cvtColor(src, cv2.COLOR_RGB2GRAY)
canny = cv2.Canny(cal_img, 200, 200)
return canny.var()
def getHogFeature(self, img:cv2.Mat, winSize=(64,128), blockSize=(16,16), blockStride=(8,8), cellSize=(8,8), nbins=9):
hog = cv2.HOGDescriptor(winSize,blockSize,blockStride,cellSize,nbins)
cal_img = cv2.cvtColor(cv2.resize(img, winSize), cv2.COLOR_BGR2GRAY)
hist = hog.compute(cal_img)
return hist
class segment:
def __init__(self, frames:list[Frame]) -> None:
# assert len(frames)>=16
self.totDate:list[Frame] = frames
self.frames = []
self.clear = 0.0
self.score1 = 0.0
self.score2 = 0.0
self.feature = []
for fr in frames:
self.frames.append(fr.frame)
self.clear+=fr.clear_score
self.feature.append(fr.hog)
self.score1 += fr.symmetry_score
self.score2 += fr.color_score
self.clear/=float(len(self.frames)) # 视频片段的清晰度得分是每帧的平均值
self.feature = np.average(np.array(self.feature), axis=0) # 视频片段的特征是每帧的平均值
self.score1/=float(len(self.frames))
self.score2/=float(len(self.frames))
# print(self.feature.shape)
# model = C3D_feature().eval()
# input_size = (224,112)
# input = [cv2.resize(i, input_size) for i in self.frames]
# input = torch.Tensor(input)
# input = torch.transpose(input, -1, 1).unsqueeze(0).transpose(1,2)
# self.c3dFeature = model(input).reshape(-1)
# model = torchvision.models.mobilenet_v2(pretrained=True).eval()
# input_size = (224,112)
# input = [cv2.resize(i, input_size) for i in self.frames]
# input = torch.Tensor(input)
# input = torch.transpose(input, -1, 1)
# self.mbFeature = model(input)
# print(self.mbFeature.shape)
def __len__(self):
return len(self.frames)
def hog_dist(self, obj):
return cv2.norm(self.feature - obj.feature, 2)
class FrameOrg:
def __init__(self, frame:cv2.Mat, idx=-1) -> None:
self.frame = frame
# self.clear_score = self.getClearFeature(frame)
# print(f"clear: {self.clear_score}")
self.symmetry_score = self.getSymmetry(frame, 1)+self.getSymmetry(frame, 0)
# print(f"symmetry:{self.symmetry_score}")
# self.symmetry1_score = self.getSymmetry(frame, 1)
# self.symmetry0_score = self.getSymmetry(frame, 0)
# self.hog = self.getHogFeature(frame)
self.color_score = self.getColor(frame)
# print(f"color: {self.color_score}")
# self.pos = 0 # For aesthetic：最终摘要中的第几帧
# self.belongWhichSegment = 0 # For aesthetic: 属于第几个片段
# self.raw_id = idx # 原始视频中的第几帧，用于实验评分
def getSymmetry(self, img:cv2.Mat, code:int):
img_cal = img.copy()
# 创建对称图
img_flip_src = cv2.flip(img_cal,code)
# 将特征点对称
# 创建sift计算子
feature_point_size = 100
sift = cv2.SIFT_create(feature_point_size)
img_gray = cv2.cvtColor(img_cal, cv2.COLOR_BGR2GRAY)
img_flip_gray = cv2.cvtColor(img_flip_src, cv2.COLOR_BGR2GRAY)
# 获取该图像的特征点
sift_src=sift.detectAndCompute(img_gray,None)
sift_flip=sift.detectAndCompute(img_flip_gray,None)
avg_sc = 0.0
theta = 50.0
for src_id, src_p in enumerate(sift_src[0]):
x,y= src_p.pt[0], src_p.pt[1]
dist = []
simi = []
for flip_id, flip_p in enumerate(sift_flip[0]):
i,j = flip_p.pt[0], flip_p.pt[1]
dis = get_dist(x,y,i,j)
if dis <= theta:
dist.append(dis)
simi.append(get_simi(sift_src[1][src_id],sift_flip[1][flip_id]))
if len(dist) != 0:
dist = torch.log_softmax(torch.Tensor(dist),dim=0)
simi = torch.log(torch.Tensor(simi))
sc = torch.max(simi-dist)
avg_sc += sc/ float(feature_point_size)
return avg_sc
def getColor(self, img:cv2.Mat):
img_src = img.copy()
img_hsv = cv2.cvtColor(img_src,cv2.COLOR_BGR2HSV)
score = 0.0
cnt =0
for i,row in enumerate(img_hsv):
for j,pixel in enumerate(row):
cnt+=1
sc = (float(pixel[1])*0.5+float(pixel[2]))*0.5/255.0
score += sc*100.0
return score/(img.shape[0]*img.shape[1])
def getClearFeature(self, src):
cal_img = cv2.cvtColor(src, cv2.COLOR_RGB2GRAY)
canny = cv2.Canny(cal_img, 200, 200)
return canny.var()
def getHogFeature(self, img:cv2.Mat, winSize=(64,128), blockSize=(16,16), blockStride=(8,8), cellSize=(8,8), nbins=9):
hog = cv2.HOGDescriptor(winSize,blockSize,blockStride,cellSize,nbins)
cal_img = cv2.cvtColor(cv2.resize(img, winSize), cv2.COLOR_BGR2GRAY)
hist = hog.compute(cal_img)
return hist
if __name__=="__main__":
import cv2
import os
from segment import FrameOrg,Frame
import matplotlib.pyplot as plt
import time
dir_path = "img\\neg"
picture_path = os.listdir(dir_path)
st = time.time()
score = []
for pic_file in picture_path:
img = cv2.imread(os.path.join(dir_path,pic_file))
frame = FrameOrg(img)
print(f"15张图片的计算时间（轻量化前）：{time.time()-st}s")
st = time.time()
score = []
for pic_file in picture_path:
img = cv2.imread(os.path.join(dir_path,pic_file))
frame = Frame(img)
print(f"15张图片的计算时间（轻量化前）：{time.time()-st}s")